SU‐FF‐T‐08: Morphological Seed Identification and Removal of the Post‐Implant Prostate Brachytherapy Patients in Cone‐Beam CT Sinogram Projections

Purpose: To digitally extract and remove elongated seed features from measured 2D cone‐beam CT (CBCT) sinogram projections of the post‐implant prostate brachytherapy patients. Methods and materials: The method identifies the seeds at the projections and removes them by 2D spatial‐interpolation. Theragenics Model‐200 [formula omitted] seed images were identified and removed from 2D patient projections by (a) normalizing the image intensity by finding its maximum and minimum values in the image, (b) top‐hat‐filtering, (c) thresholding using the 3σ‐value of the pixel intensity histogram, and (d) labeling to create a binary mask of each seed in each projection. The binary masks were subtracted from the corresponding raw projection. The subtracted images were then interpolated in the seed regions using natural‐neighbor interpolation method to recover the missing soft‐tissues information's obscured by the implanted seed images. Four [formula omitted] post‐implant patients were scanned using an Acuity‐digital‐simulator with full‐660 projections CBCT (in half/full‐fan‐mode) for post‐implant dosimetry. In‐house filtered‐back projection algorithm was used to reconstruct CBCT images before and after seed removal applications. To quantify this method, we compared the reconstructed CBCT images before and after corrections, image‐profiles and a difference image on a single central slice to demonstrate the effects of seeds removal. Results: For the example cases, the streaking artifact is reduced by a factor of 3.5 and artifact spatial extend. The corrected images exhibit significantly improved image quality in and around the prostate. Discussion: Our preliminary results indicate that this method can be used to mitigate metal‐streaking artifacts specific to the brachytherapy seed implant geometry. By reducing streak and associated noise propagation artifacts, significant clinical value can be added to brachytherapy CBCT imaging. Improving the auto‐segmentation method and applying scatter‐subtraction corrections will be further study to improve the CBCT image quality for the intra/postoperative brachytherapy patient's images.